ToolsConductScience tool
Lander–WatermanFree in-browser calculator

NGS Read Coverage Calculator.

Reverse-solve reads-per-sample, samples-per-run, runs-needed, and cost-per-sample for any sequencing project. Compare Illumina NovaSeq X / NextSeq 2000 / MiSeq, Element AVITI, Oxford Nanopore PromethION, and PacBio Revio side-by-side. Lander–Waterman math, runs entirely in your browser.

PrivateData stays in your browser
LiveNo sign-up required
Open methodFormula shown on this page
CitableMethods and citation included

Calculator

Results update in place

Try it out

Load example NGS coverage data to see the full workflow

bp
×
n
bp
USD

Coverage Plan — Illumina NovaSeq X — 25B (2×150 PE)

Per sample

Reads / sample
320.0M
Yield / sample
96.00 Gb
Coverage
30×
Cost / sample
$192.00

Project totals — 10 samples

Samples / run
81
Runs needed
1
Total yield
960.00 Gb
Total cost
$1,920

Comparison across platforms (3.20 Gbp target, 30× coverage, 10 samples)

PlatformReads / sampleYield / sampleSamples / runRunsCost / sample
Illumina NovaSeq X — 25B (2×150 PE)320.0M96.00 Gb811$192.00
Illumina NovaSeq X — 10B (2×150 PE)320.0M96.00 Gb321$192.00
Illumina NovaSeq X — 1.5B (2×150 PE)320.0M96.00 Gb52$192.00
Illumina NextSeq 2000 — P3 (2×150 PE)320.0M96.00 Gb33$192.00
Illumina NextSeq 2000 — P2 (2×150 PE)320.0M96.00 Gb18$192.00
Illumina NextSeq 2000 — P1 (2×150 PE)320.0M96.00 Gb132$192.00
Illumina MiSeq — v3 (2×300 PE)160.0M96.00 Gb164$192.00
Illumina MiSeq — v2 (2×250 PE)192.0M96.00 Gb1128$192.00
Element AVITI — High Output (2×300 PE)160.0M96.00 Gb62$192.00
Element AVITI — High Output (2×150 PE)320.0M96.00 Gb34$192.00
Element AVITI — High Output (2×75 PE)640.0M96.00 Gb17$192.00
ONT PromethION — single R10.4.1 flow cell9.6M96.00 Gb17$192.00
PacBio Revio — single SMRT Cell HiFi5.3M96.00 Gb111$192.00

Cost-per-sample uses your single cost-per-Gb input across all platforms — adjust per platform for an accurate quote.

When to use

  • Plan reads-per-sample, samples-per-run, and runs-needed for any NGS project from genome size and target coverage
  • Compare Illumina, Element, ONT, and PacBio platforms side-by-side for the same project
  • Estimate cost-per-sample given a quoted cost-per-Gb from your core facility
  • Reverse-solve coverage from a fixed read budget (e.g. amplicon at 50k reads / sample, single-cell at 50k reads / cell)
  • Sanity-check whether a project will fit on a single run or needs to be split across runs

Do not use for

  • Variant-call sensitivity prediction — use a per-base power calculator (DeepVariant / GATK CalculateContamination) for low-AF detection
  • Modelling Poisson coverage tails or coverage-uniformity issues from capture / GC bias
  • Long-read assembly contig N50 prediction — depth alone does not predict assembly quality
  • Single-cell experiments where you need *per-cell* sensitivity at a specific gene — coverage on a transcriptome is the wrong abstraction

Paired-end is L = 2 × read length

A 2×150 PE run delivers 300 bp per read pair, not 150. The calculator stores combined read length so the formula stays N = C·G / L without per-platform special-cases. If you switch to a single-end mode for a long-read run, halve L manually.

Headline yield is nominal, not guaranteed

Manufacturer headline yields are top-of-spec. Cluster density variability, library quality, and run failures typically take real-world yield to 80–100% of nominal. Build in 20% slack for production runs.

Coverage is an average, not a floor

Reads land randomly. ~5% of bases sit below half-coverage. For confident germline variant calling on diploids, plan 2–3× the headline depth so the Poisson tail still has enough per-allele reads.

Multiplexing has a floor

You cannot put 1,000 samples on one MiSeq just because the math says reads-per-sample fits. Index well-balance, well-to-well leakage, and minimum reads-for-demultiplexing all set practical floors. For most Illumina platforms, do not exceed 96-plex without index-hopping mitigation.

WES coverage ≠ WGS coverage

100× WES is roughly 6× WGS in raw bases — not because the math is different, but because the target is 50× smaller. Variant calling sensitivity at 100× WES on coding sequence is roughly equivalent to 30× WGS. Do not benchmark them as if they were the same metric.

1

Method

Lander–Waterman coverage relation C = L·N / G, reverse-solved for reads-per-sample, samples-per-run, runs-needed, total yield, and cost-per-sample. Platform throughputs are manufacturer-published headline values from current spec sheets (Illumina, Element Biosciences, Oxford Nanopore, PacBio). Application defaults are drawn from GATK Best Practices, ENCODE guidelines, the Earth Microbiome Project, 10x Genomics user guides, and the NIH Roadmap Epigenomics consortium.

2

Published

Published 2026-04-07. Calculations support planning and documentation; verify procurement decisions against manufacturer specifications or institutional SOPs.

3

How to cite

How to Cite

ConductScience NGS Read Coverage Calculator. ConductScience, Inc. 2026. Available at: https://conductscience.com/tools/ngs-read-coverage-calculator

Lander ES, Waterman MS. Genomic mapping by fingerprinting random clones: a mathematical analysis. Genomics. 1988;2(3):231-239. doi:10.1016/0888-7543(88)90007-9

Sims D, Sudbery I, Ilott NE, Heger A, Ponting CP. Sequencing depth and coverage: key considerations in genomic analyses. Nat Rev Genet. 2014;15(2):121-132. doi:10.1038/nrg3642

Van der Auwera GA, et al. From FastQ Data to High-Confidence Variant Calls: GATK Best Practices Pipeline. Curr Protoc Bioinformatics. 2013;43:11.10.1-11.10.33. doi:10.1002/0471250953.bi1110s43

What is read coverage?

Read coverage (also called sequencing depth) is the average number of reads that overlap each base in your reference. A 30× whole-genome sequencing project means every base in the human genome is covered, on average, by 30 reads.

Coverage is the single most important budget knob in an NGS experiment: it sets sensitivity for variant calling, the noise floor for differential expression, and the floor for detecting low-frequency clones in tumor or microbial samples. Picking the right coverage — and the right number of samples — for your platform of choice is what this calculator is for.

The Lander–Waterman relation

For random shotgun sequencing, the expected coverage is:

C = L · N / G

where C is the average coverage depth, L is the effective read length per read or read pair, N is the number of reads (or read pairs), and G is the genome or target size in base pairs.

The relation reverses cleanly: N = C · G / L gives the reads required to hit a coverage target. Multiply by sample count and you have your project budget. Divide by reads-per-run and you have the number of runs.

Lander & Waterman (1988) derived this in the context of physical mapping; the formula carries over directly to sequencing because both processes are random sampling on a target.

Picking a platform

Population-scale WGS / large cohorts: Illumina NovaSeq X 25B is the default — ~8 Tb / run lets you multiplex dozens of human WGS samples per run at the lowest cost-per-sample on the market.
Mid-scale WGS / WES / RNA-Seq: NextSeq 2000 P3 (~360 Gb / run) is the workhorse for one-off projects of a few to a few dozen samples.
Targeted panels and amplicon work: MiSeq v2/v3 — small batches, fast turnaround, paired-end 2×300 for full V3-V4 16S amplicon coverage.
Cost-sensitive Illumina alternative: Element AVITI delivers similar 2×150 throughput to NextSeq 2000 P3 at competitive cost-per-Gb, with no Illumina lock-in.
Long reads (structural variants, full-length isoforms, de novo assembly): ONT PromethION (real-time, ~10 kb mean) or PacBio Revio (~18 kb HiFi at QV30+). These complement short reads — they are not a substitute when you need accurate variant calls.

Frequently asked

What math does this calculator use?
It uses the Lander–Waterman coverage relation: C = L·N / G, where C is the average coverage depth, L is the effective read length per read or read pair, N is the number of reads (or read pairs for paired-end), and G is the genome / target size in base pairs. The same formula reverse-solves for reads needed to hit a coverage target: N = C·G / L. This is the relation behind every standard NGS coverage calculator (NEB, Illumina, Element).
How are paired-end reads counted?
For paired-end runs, the calculator stores the *combined* read length (e.g. 2×150 PE → L = 300 bp) and counts read pairs as N. So a 30× human WGS on NovaSeq X 25B (2×150 PE) needs ≈320M read pairs, which yields ≈96 Gb of sequencing per sample. This matches Illumina’s own coverage tables.
Where do the platform throughput numbers come from?
They are manufacturer-published headline yields from current spec sheets: Illumina NovaSeq X (M-GL-01218), NextSeq 2000 (M-GL-00620), MiSeq (M-GL-01233), Element Biosciences AVITI System Specifications, Oxford Nanopore PromethION product brief, and PacBio Revio (101-871-100). Real-world yield is usually 80–100% of headline; treat the calculator output as nominal capacity.
Why is my Whole Exome project so much smaller than my WGS project?
Because the *target* is smaller. WES captures ~60 Mb of coding sequence — about 50× less DNA than the 3.2 Gb human genome. Even at 100× depth, WES needs only ~20M read pairs vs ~320M for 30× WGS. The trade-off: WES is cheaper per sample but blind to the 99% of the genome outside the capture.
What coverage should I aim for?
Sensible defaults: 30× for human germline WGS (GATK best practices), 100× for WES (deeper for somatic / tumor), 30M read-pairs / sample for bulk RNA-Seq differential expression (ENCODE), 20M unique reads / sample for sharp ChIP marks, ≥45M for broad histone marks, 50k reads / cell for single-cell 3′ on 10x, ≥30× for whole-genome bisulfite sequencing. Override these with your sensitivity requirements.
How realistic is the cost-per-sample number?
It is yield_per_sample ×\times your cost_per_Gb, nothing more. The default cost_per_Gb of 2reflectscurrentspotpricesforNovaSeqXreagentsinearly2026.Youractualcostwillincludelibraryprep(2 reflects current spot prices for NovaSeq X reagents in early 2026. Your actual cost will include library prep (50–$200 / sample), capture / depletion reagents, instrument time, labour, and overhead. Use the per-sample yield as the variable cost and add a fixed overhead from your core facility quote.
Why does Lander–Waterman give average coverage, not minimum?
Because reads land randomly across the genome, the actual coverage at any base follows a Poisson distribution with mean C. About 5% of bases will be covered at less than ~½C, and about 1% at less than ~⅓C. For confident variant calling on diploid genomes, plan for 2–3× the headline coverage so that the Poisson tail still gives you enough per-allele depth.
Does my data leave the browser?
No. All computation runs locally in JavaScript. Nothing is sent to any server. You can verify by disconnecting from the internet and reloading the tool.
How should I cite this tool?
See the "How to cite" section on this page for a ready-to-paste methods statement. The underlying math is from Lander & Waterman (Genomics, 1988); platform throughputs are from manufacturer spec sheets.
325
Free tools
1,200+
Institutions
100%
Client-side
0
Uploads required